Breaking barriers in neurotechnology: The future of brain-computer interfaces

0:01It's been such a good collaboration, but it did require literally just like, I'm here. and like prior to that it was like the week before it was like oh what are we gonna do like we have no collaboration now and that took like months to get in place no way yeah it was really bad um so the full circle of that is like it went like catastrophically bad and then everyone's like wow you have a collaboration with Stanford it's like well actually here's what really happened and then we like yeah obviously where we're at now is a very different story and we have an awesome collaboration with them now.

0:40Welcome back to Wild Hearts. I'm your host, Mason Yates. I'm on the investment team at Blackbird Ventures. Today, we're joined by Dr. Elise Jenkins, a visionary scientist and relentless builder at the frontier of neuroscience and oncology. Elise and the team at OptoBio are pioneering breakthroughs in brain-computer interfaces, tiny implantable devices that don't just monitor brain activity, but they dream of transforming how we diagnose and treat cancer. Elise's journey has taken her from concept to the cusp of groundbreaking human trials, involving everything from navigating intense engineering challenges like implanting devices invisible to MRR, to forging collaborations with some of the world's leading researchers by rocking up to their doors.

1:23And I mean that literally. In this episode, Elise shares candid reflections on her journey from academia into the startup world, revealing the critical decisions, unexpected setbacks, and the intelligent ways she's dealt with the constraints which have propelled her company forward. Prepare yourselves to learn about the intersection of neuroscience, oncology, and the relentless progress towards human trials. Let's jump in.

1:50Yeah, I would struggle to zoom out, but if you can zoom out and summarize the last 12 months, that'd be unbelievable. Could I go a little bit further back to like the last 18 months um why 18 months because that was kind of like how long our last round was and that like there was a lot that sort of happened to to get us to where we got to in there in the first round yeah you know about sort of january february 2023 i think it was nearly two years ago now um was when we closed we started actually building on our first close of the first round um and so we started pulling together a team of i guess experts from the pre-clinical space all the way through to people who have developed implants all the way through from preclinical translation and human trials.

2:35And so over the last 18 months, we sort of set out to build the smallest brain computer interface that currently exists both in pre-market and currently in market. And basically develop an MVP of that interface, show that it was safe, show that we could work with clinical neurosurgeons, implant this in different types of models, and build together the team to actually to do that as well. So that's kind of been our focus for the last 18 months. Alongside that, building that MVP out, also really looking at our first indication, how are we going to use this and how are we going to bring benefit to patient populations?

3:09For us, that really has been a massive focus on getting what preclinical information do we want to really understand so we understand our target indication. And that has been focused on oncology, so a very new application for these types of interfaces, in particular brain tumors. So we spent a lot of time working with a group in Stanford with Michelle Mongey, working on how can we use neural interfaces or brain-computer interfaces in the application of oncology, in particular in brain tumors. So how do we use electrical information from the brain? How can we understand how this varies and changes over time with disease present?

3:46And how can we use that information now as a biomarker to tell us how a patient's disease might be progressing? Are they responding to their treatment? How can we tell that by the day rather than waiting six months, nine months until a patient has an MRI? And so that was like a massive milestone for us, was to develop the first neural biomarker, first indicator of disease progression just by looking. When was that milestone? When? Yeah. I'm just like, we raised, built the MVP and holy moly, we just showed we have the first neurobiomarker? Confirmation of that took about 12 to 18 months. So we built an MVP and we also got that data in 18 months from first race, from basically concept to first.

4:28What was surprising about the results? There was a few things. We had a hunch, right? I mean, I've had a hunch for maybe six years now that from my research in my PhD, I worked alongside a lot of people who did neuroscience and neurotechnology research and development. And then my application or my focus and my work was in brain tumors. And there was this massive disconnect between what people do in neuroscience and what people do in cancer biology. And I felt like this really lucky person that had both in my visual and could see the overlap. And so that was kind of what really led us to doing a lot of this work was that there was this big hunch that if you use tools and use technologies from neuroscience and you apply them in cancer, you will see that you can basically use all of this information to understand how cancer progresses and also how we can block it and how we can slow it down.

5:19So I had this hunch years ago that you should be able to make this, this should happen. People had already shown that there was a difference like in a snapshot. So if you looked at a tumor brain and you looked at a healthy brain, there was a difference in electrical activity, but no one knew that it changed over time. And that was kind of our hunch that that would happen because the brain is very plastic and actually rewires quite a lot during disease. So being able to look at that progression was very significant. And so there were no prior scans that showed different activity over time with a patient.

5:54That seems bizarre to me. Yeah. Yeah. This overlap of fields is only really coming to fruition in the last five years. Whoa. Yeah. I'm amazed. Why do you think that is? Oh, man. This is probably like a more generalist versus specialist argument and more philosophical. But I think research over time has become so siloed. And so you have, it goes siloed and then it kind of expands again and goes siloed. And I think we were in a period for a while where research became very individualized. People were looking at very specific ions or protein channels, sorry, ion channels or proteins in cancer biology.

6:32And there was collaboration, but it was collaboration with the same people who were doing these types of very specific research. And similarly in neuroscience, right? Neuroscience, people are looking at how does the brain evolve? how we have circuit dynamic changes in the brain, what does that mean for specific disease. But cancer has never been associated as an electrical disease. It's very much an electrical disease. And that's starting to become very apparent now. And so this field is very much exploding where you have cancer and neuroscientists finally working together. And I think I was probably one of the lucky few who knew both and could see very much like, well, we know that most cells in the body are very electrically active.

7:10It's very, very true for cancer cells. We've studied neuroscience for decades using these tools, looking at electrical activity in brain cells, just kind of drew the dots. And we're like, well, why don't we use those here? What does it mean? And it means a lot. And now people, like there are whole fields, there are lots of bodies of funding that are going towards this space now. And we're right at the forefront, which is awesome. That is amazing. And maybe go a bit deeper on, there's one side of identifying the activity, and then there's the other side of actually stimulating it and changing its shape.

7:46Can you share a bit more about that? Where we got the idea to look at, can you use an electrode to record information in the brain? And what does that have? What impact would that have if there is a cancer present? Was fundamentally about this discovery where probably in the last decade, people have identified that neurons in the brain, so your nerve cells that are responsible for sending messages, not only form a network with the tumor, but they actually form what's called a functional synapse with cancer cells. This discovery is very important and very remarkable and very scary because that means that there is a bi-directional communication that happens now between neurons in the brain and cancer cells.

8:28And that communication is essentially seen as growth signals for cancers. So cancers release certain factors into the environment that recruit neurons into them. And then once they have them, they then harness them to essentially grow very, very fast. And because we know that, obviously, from neuroscience, you can record all of this information from neurons. So the hunch was, well, if that's happening with neurons and it's happening with cancer, then you can record that activity. And so that was what we did first. And we're like, okay, let's see how that changes through time. Very, very consistent week to week variation in brain activity.

9:05There are very specific frequencies of information. These are called high gamma sort of bands of frequency information that progress through time that tells us that disease is progressing. And then very much in a very similar fashion, we know from neuroscience that you can modulate neurons either pharmacologically, electrically. There are a multitude of ways you can modulate behavior of neurons to slow them down or stop them from communicating in certain ways or make them communicate more depending on what disease indication you're looking at. And so very, very similarly from that, now our hunch is, well, why don't we not try and modulate that and see how we can slow those interactions down to slow down the tumor growth?

9:47And so that's a lot of the work that we've been doing now has looked at that. From a pharmacological point of view, people block these activities, these communication signals happen between neurons and cancer cells by using actually anti-epileptics are a common drug that people have been trying to do. essentially silencing some of those interactions, which show a significant increase in survival in preclinical models for cancer. People have used different forms of electrical stimulation to try and target cancer cells. So there is a known phenomenon, a known, I guess, survival benefit for using electrical stimulation.

10:20And just we're looking at how can we optimize some of those strategies and what type of impact could we look at in terms of survival in patients. and maybe share a bit more about the process that you went on to validate the results what sort of data set did you need to get to that conviction and how did you acquire that data set yeah so um a lot of our work at the moment is done in preclinical models so um these are typically done in mice um we use a very well characterized model with uh research groups uh so um the group we work with in Stanford have an incredible model that we can use to, that's like very well published and well characterized that we can use to essentially take patient samples from tumors and allow them to grow in mice models in the brain that we would typically see these types of tumors exist.

11:17And essentially what we do is you run longitudinal studies. So for recording data, so looking at biomarkers, these are anywhere from two to four months long the study lasts. And depending on what you're trying to see, you essentially look at how you implant these devices into the brain. We look in the premotor cortex, which is a very elegant area of brain where typically a lot of tumors do present. And we essentially look at recording information in a freely moving sort of awake environment for these animals. and we just record electrical activity for a period of anywhere between 10 minutes and an hour each for each animal you do this across many so that you can cross validate using different machine learning techniques and then essentially you take that information each week up until a specific time point of interest and then you correlate that against the histology so how big is the tumor when you actually you know look at the end of this and so you look at how your week to week variation changes and how does that compare to histological or other methods like there's a method of bioluminescence where you can actually measure the size of the tumor as well.

12:27What does that word mean? Bioluminescence. Yeah I couldn't even repeat it. It's essentially a way that you tag cells with a fluorescent indicator and so you put them into a specific machine you expose it to a specific wavelength and then they will illuminate and the amount of illumination is relevant to how big the tumor is. And can you go deeper on the specific neurobiomarkers that you're tracking? Yes. So these are electrical information. So when you look at, the easiest way to kind of explain this is if you can relate it to something like a seizure. And so when people have seizure-like activity and you put electrodes on the head, these are EEG type of electrodes, what you see is normal brain behavior looks in a certain way.

13:20You'll have certain spikes or certain types of waveforms that are considered healthy and normal. And then in a seizure, when a patient is having a seizure, you start to see very sudden bursts of activity that look abnormal. And while we don't see that type of birth-like behavior in cancer, what we do see is that there is abnormal hyper-excited regions in the brain that are progressively getting more hyper-excitable. So rather than seeing sort of like births and recovery from a seizure, we're kind of just seeing this activity getting more and more and more powerful over the duration of this disease.

14:00We see this in two specific bands in the brain that are very commonly studied. Some of these are in high-frequency activity. So this is very common for or commonly represented as spike-like activity or hyper-excited regions of the brain. Lots of activity happening at once. And we also see this in very, very slow wave activities. So this is called the delta range. We also see variance or week-to-week variation in very, very low-frequency activity. So that's how we've been looking at the biomarkers in the brain. If we rewind back to before you raised the last round, 18 months, what did you need to show in order to raise the round?

14:42And how did you build conviction that there was like, oh, we're now ready? Yeah, so Ben and I were in our PhDs when we were raising our first round. We were very much raising on more of a concept than any validated MVP. What we had done in the lead up to the first round was we had actually done more on a new concept of device rather than validation of specific biomarkers. So Ben and I had worked specifically on glioma from a neural interfacing point of view. And Ben had worked on making very, very small devices for the spinal cord, looking at spinal cord pain and rehabilitation. And we spoke about maybe for about a year before we actually raised, where we had come together, come up with these concepts to make very, very, very teeny tiny devices.

15:37And our idea at the time was that you would make these very small, essentially transistor-based sensors that we would couple with light and we would scatter them on the brain and we would make what's known in our world as an optical neural dust. That's essentially what we were trying to create. Little tiny devices, scatter them on the brain and the brain will light up when we're trying to record information from them. This was a very wild idea. We had done some proof of concept. We filed a bit of IP on the actual concept and done some very, very early characterization of that. And when we went out to, you know, raise from Blackbird, for example, we really wanted to take that concept and make an actual technology device, make an actual MVP of that and show that we could make it safe, show that we could implant it in, work with neurosurgeons, show that we could implant that in sort of human cadaveric models.

16:33and really prove out our first indication at space, which was, is there an application here in brain cancer? That was a sort of bad question. So that was what we had done in the lead up. Interesting. And then how did you arrive on the UX? Like why did you need an implantable device versus a headband, for example? There's a number of reasons why you would, it depends on your application to start off with, But there are a number of reasons why an implantable makes sense as a medical device for a medical indication. And we're talking about, you know, looking at these types of interfaces or wearable technologies that are helping you perhaps focus more or, you know, wanting to improve mood in these different sort of application spaces or more consumer facing applications.

17:24It's very hard to convince anyone that they should get a brain implant to have something like that. and the need, the medical need is not necessarily there to warrant an invasive surgery. And so for a lot of medical needs, sort of we're talking about cancer, epilepsy, Parkinson's, depression, where the need is very, very much there. There's very much a clinical need and the risk of surgery is perhaps they're already doing a surgery, which is in our case, very helpful. Or the idea to warrant the use of implanting a device that is continually there, you kind of set and forget is the, I guess, the principle.

18:01Those comes down to the, I guess, the user needs. But the second part to that is actually technically what is feasible. So if you're wanting to read low frequency information, so information like, am I blinking? Am I concentrating? What is my mood? For example, you don't need high frequency information. And so you can get a lot of that information from outside of the head, albeit it is noisy, you can get it. When you're talking about high frequency information, high fidelity information to help diagnose a medical condition or something where intervention is very, very, you need high fidelity information for something like that, you need to be in the brain.

18:41So I haven't seen a single technology where an electrode is able to pick up high frequency information outside of the skull. You have a filter there that stops all of that information coming out. So it depends on your application and your need, but for our application and our need, it needs to be in the brain. We need that information and we're directly, we need to be directly interfacing with the brain tissue to deliver the types of stimulus that we need at the specificity that we require. So it really depends on the application and need. And what did you learn about the milestones or units of progress that you needed to accomplish having raised that round coming from a background of academia to a background, like into a world of startups?

19:21I mean, I had an industry break between academia. I went to industry and then I did PhD and then came back to startup. But I think some of the things that I, that may have surprised me a little bit or what we learned is that one, sometimes the smallest thing, like we're building the coolest, smallest engineering thing isn't perhaps as useful to clinicians. And so really trying to understand what is your patient need or what is the need that you're actually solving and build to that and try not to over-engineer things, which as an engineer is very hard to do and accept. But so that was a big one.

19:55One of the things that we went out to, we were like, we're going to scatter these devices over the brain. And we're talking to these neurosurgeons. They're like, how do you get it in? How do you get it out? And we were like, what's the point?

20:08So there's a lot of learnings there that were like, there is such a thing as too small. You really need to think about human factors and how they would interface with this. And that really also came from speaking from neurosurgeons, but also bringing in med device experts. So people who have built these systems, made them robust, get them into patients. I think with that is another learning that from academia to making a product is like, okay, in academia, how many times do you have to get it to work to get your published paper? And a lot of people are not really necessarily caring about the translation.

20:41They're like, okay, well, I need this so I can go to my next academic role or whatever it might be, which is very sad. And that was part of the reason why I wanted to leave. Whereas when you're translating a device, it can't just work three times to get the three repeats of your paper. It has to work every single time. And so again, bringing the right people in with that framing of mind to get our engineering team and our science working every single time. It can't work a few times. It has to work every time. And that requires a lot more thought in the design, a lot more thinking in the development scheme.

21:13And so that's been a massive learning and challenge, I think, throughout the last year and a half. Making things a bit bigger to suit the standard of care and to suit the clinical application, but also making it work every time is, I think, a feat that academics may disregard. Can you go deeper on the second one, especially as it relates to an example that comes to mind on making sure that it is bomb-proof? Yeah. I'm going to use Neuralink as an example because it's the easiest one that people, whoever's listening to this will understand. But when Neuralink made their announcements about the types of interface, like neural interfaces that they were developing, the academic world kicked off and everyone was like, people have been doing this for decades.

21:55You know, we've been doing this in research for decades. We can, like these materials, there's nothing novel that you're doing here. And I was part of that crowd initially. I was like, oh, they're not doing anything very interesting. We've been doing this forever and they're all getting all this hype for it. And then when we started building our company, I realized it's very, very hard to make something work 100 % of the time. And there's a big difference between publishing a paper and getting something to work in the lab without all these controls in place to actually getting it to work in a human grade every time and safely.

22:29And so I now have a massive appreciation for the difference between novelty and getting something working in an academic lab to product and making something work every single time. So that's, I guess, the anecdotal reference that I can use there. And again, it's been like how you document things, how you ensure that you have the right controls in place to ensure that the way that you build something once, you then build it again and it's the exact same. There can't be a difference in how you've built it. getting myself familiar with that process, working with the right people to ensure we do that properly, building out our team's capabilities to do that.

23:04It's definitely been a learning curve for us, but it's been awesome. It's really good for us. Can you entertain us with some of the high level trade-offs that you need to make? From an engineering point of view? Yeah. Health, safety, long-term roadmap, being bomb-proof, the list goes on. What are some of the non-negotiables that have to be a part of that MVP? And what were some things that kept you up at night and ultimately made a different route decision? There are a lot of things that came up as we understood our patient population better. And we spoke more with our neurosurgeons about what is the actual need and what are the sort of parameters that we can work within.

23:43And an absolute non-negotiable that came up is this device. Not only does it have to be MRI safe, so not only does it have to be able to go through an MRI machine, but it can't cause an artifact. So, you know, if you think about a deep brain stimulator or, you know, any sort of device that's implanted today, include Neuralink in this, Synchron is the same. If you think about any of these implants, you know, there'll be, some of them may be classed as MRI safe. You can go in and it's not going to rip out of the brain in the middle of an MRI, but they'll cast a shadow. Like the way an MRI machine works will essentially absorb certain materials, the metals will absorb some of the RF field.

24:23It will create a black shadow around the implant and then you can't see tissue. Now for brain tumors, the only way you can diagnose a lot of these is through an MRI or you can monitor their progression. So a non-negotiable for us is that not only can you do the procedure, like do an MRI procedure and it doesn't cause a problem, but you need to make it as transparent as possible. This is an engineering problem because everything's made out of metal right there's metal in everything yeah and so we've had a massive engineering challenge but also a very exciting challenge to make an mri invisible implant um which has involved taking out every single bit as we possibly can making everything out of like non-absorbent materials uh non-rf absorbent materials so uh you know our interface that goes into the brain these are made from polymer materials on thin film metals, which you can't see on the MR.

25:20That was something we were already doing, so that wasn't anything too difficult. But the piece which holds all the electronics sits in the skull. And so making everything out of glass, removing the battery, trying to essentially use ceramics and different materials to not absorb the RF has been, but also it's a wireless influence so there is an rf coil in there how much does that yes there's been lots of challenges there but um so far um we've essentially managed to do a lot of interesting engineering to to make this device out of mostly out of class so that's been a big engineering thing i think a lot of fun for the engineers on the team yeah that is seriously wild and fun to think about you asked um before about the like you know trade-offs um i think one thing that comes up a lot comes up a lot between companies.

26:12It comes up a lot as like a metric equivalent to like when people are talking about quantum. They're like, well, how many qubits do you have? You know, a question that comes up a lot in our space is, well, how many channels do you have? How many electrodes do you have interfacing with the brain? And well, this company has 64 and this one has 12 and this one's got 1024. And I think one of the trade-offs that you have to make at some point is what is the density of those channels and how are you actually going to transmit that data route so when you have a an interface where you're trying to make it very small you have to trade off size with essentially memory because which then affects your channels so if you start having increasing your channel count you you're sampling everything at the same time if you don't have a high enough data rate out you have to store it if you have to store it you need to have space on your implant to do that which makes your footprint bigger.

27:06So as you start to increase those types of channels and you hit the bandwidth limitations of most wireless communication systems, you then need to try and store this data in some capacity, which then increases your footprint. So trying to balance size and channel count and data rate outs has definitely been another challenge that we've been trying to balance through the last year and a half. Why does channel count matter, especially when you're this focused on a particular problem? It's a great question. I wish most people thought of it like that. It's a vanity metric, I think, a lot of the time.